What an AI-driven claimant like Morgan & Morgan means for your defence and evidence strategy

You must document which AI systems you use for evidence and risk analysis, record the controls and logs that exist per workflow, and ensure your own AI output remains verifiable. Your governance must keep pace with the AI capability on the opposing side.

An analysis of 14 September 2026 of AI-driven case selection and evidence handling at scale in claimant firms argues that external capital flowing into litigation technology is being explicitly linked to scaling AI and legal tech within claimant practices. Morgan & Morgan, the largest personal-injury firm in the United States, has engaged JPMorgan to explore a minority stake that could yield more than one billion dollars, with investors attracted precisely by the ability to deploy technology and AI at scale. In our assessment, the relevance for defence teams, insurers and public organisations is that claimant firms are becoming hybrid entities—part legal practice, part data-driven claims organisation—and opposing parties must now account for their own AI use at a comparable level of rigour and transparency.

What has changed about how claimant firms select and build cases?

The capital injection strengthens an existing strategy rather than introducing a new one. The firm operates deep-learning and natural-language-processing systems built for document intelligence, with automated classification and summarisation of large volumes of legal documents and predictive models for case outcomes. The appointment of a former senior Amazon engineer as the first Chief AI Officer, responsible for AI strategy across the whole organisation, signals that this use is structural and executive-level. That combination of an in-house platform, document intelligence and executive accountability turns a claimant firm into something materially different from a traditional legal practice.

What risks does this create for your own evidence and AI use?

Defence teams and insurers now face an asymmetric risk: the claimant firm can present its AI use as an organised, documented process, whilst your own use may remain undocumented and opaque. Much of your own AI use leans on models from large providers and cloud services, which makes the demand for demonstrable control sharper still. If you cannot account for which AI systems touch which workflows, what controls exist, and how outputs are verified, you sit at the table with a credibility disadvantage.

The failure modes and risks at issue are:

• Undocumented AI intake — case selection, scoring and vetting processes that use AI but leave no audit trail of how decisions were made.
• Unverifiable evidence summaries — AI-generated summaries of documents or case files presented without visible substantiation or source attribution.
• Uncontrolled model outputs — use of large language models or other AI systems without intermediate verification or correction steps.
• Asymmetric disclosure — opposing counsel able to explain their AI governance whilst you cannot explain yours.
• Dependency on vendor assurance — reliance on a single provider's claim of correctness rather than independent verification.

Which concrete controls must you be able to demonstrate?

Setting up demonstrable AI governance is not a theoretical exercise but concrete preparation for litigation against AI-driven claimant firms. You must be able to show:

1. Record the model and its purpose — document which AI system each workflow uses, what data it processes, and the lawful basis for that processing.
2. Establish intake vetting — implement a process by which AI-assisted case selection, scoring or triage is reviewed and approved before it influences resource allocation.
3. Route outputs through independent verification — send AI summaries or analysis through a second model or human review step that makes disagreements, corrections and sources visible.
4. Preserve source attribution — ensure that every AI-generated summary or conclusion can be traced back to the original document and the specific passages it draws from.
5. Log all processing steps — maintain a record of which AI systems touched which documents, when, and what outputs were generated and used.
6. Test for hallucination and drift — periodically verify that your AI systems are not reproducing or inferring information not present in source materials.

How can you make AI output verifiable without slowing down your workflow?

Auditability begins with making visible what your own AI does, not with trusting a single answer. A verification layer can route a task through selected independent AI models and make verification steps, corrections, disagreements and sources visible for inspection. That is emphatically not a guarantee of correctness and offers no guarantee against errors in the output; it makes control possible and keeps the final judgement with the user. For confidential documents, privacy-preserving techniques can replace sensitive values before AI processing with synthetic, session-only equivalents, after which the original values are restored locally; if the privacy check fails, the document is not sent onward.

What tooling can carry and what remains your own judgement?

The core of this story lies with the capital round and the claimant firm's stance on AI, not with any single product or vendor. Tooling can make verification steps visible, route outputs through independent models, and preserve audit trails. What tooling cannot do is replace your own professional judgement about whether the evidence is sound, whether the AI output is fit for purpose, or whether the case should proceed. The lesson for your organisation is practical: make sure your internal AI governance and verification keep pace with the AI-driven litigation on the other side, so that you do not sit at the table with a credibility disadvantage.